Tamper-evident audit trails for ADK agents via asqav

[jagmarques]

Built an integration that signs every ADK tool call with ML-DSA-65 (quantum-safe). Each action gets a cryptographic signature chained to the previous one, creating a tamper-evident audit trail.

The hook sits in before/after tool callbacks:

import asqav
from asqav.extras.google_adk import AsqavCallback

asqav.init(api_key="sk_...")
callback = AsqavCallback()

agent = Agent(
    model="gemini-2.0-flash",
    before_tool_callback=callback.before_tool,
    after_tool_callback=callback.after_tool,
)

Every tool execution gets signed. The chain is exportable as JSON for compliance teams. Useful if you are building ADK agents for finance, healthcare, or any domain where EU AI Act Article 12 applies.

GitHub: https://github.com/jagmarques/asqav-sdk
Docs: https://asqav.com/docs/